Linking the inner and outer mental representations of the body to social cognition skills: A systematic review and meta-analysis

An interesting interpretation of embodiment highlights the critical role of mental body representations (BR), including motor, somatosensory, and interoceptive formats, for social cognition. However, the nature of this relationship is still debated at the empirical level, with various studies arriving at different conclusions. Here, we present a systematic review and meta-analysis aimed to summarize data from 3466 participants in 21 studies to provide a more comprehensive understanding of the relationship between inner and outer BR and social cognition. We focused on two core social cognition aspects: empathy and Theory of Mind (ToM). Concerning the inner BR, our meta-analytic findings reveal a significant correlation between specific interoceptive dimensions (i. e., interoceptive sensibility and accuracy) and social cognition, which was stronger for empathy than ToM. Conversely, although further research is needed, functional BR that mainly involve the outer body processing (i. e., nonaction-oriented BR) may show positive links with ToM. These findings point to specific interactions be-tween BR and social cognition skills, supporting multi-faceted and embodied social cognition models. However, we also identified critical knowledge gaps and highlighted the need for further investigation to deepen our theoretical understanding of these relationships and their implications for clinical practice.


Introduction
Recent neurophysiological models and experimental paradigms (e. g., the rubber hand illusion, full-body illusion, enfacement illusion, and Pinocchio illusion; see Kilteni et al., 2015), together with studies on patients with nervous system damage, suggest that mental representations of one's own body develop from the integration of different bodily sources of information (Berlucchi and Aglioti, 2010;Suzuki et al., 2013;Park and Blanke, 2019;Boccia et al., 2020a,b; see also Raimo et al., 2022).This information can be categorized into motor information and into sensory bodily information originating from the internal body, such as visceral input from the heart and the lungs (i.e., interoceptive signals), from the skin surface and the vestibular and proprioceptive senses, and from the space outside the body (e.g., visual cues involving the body and bodily sounds; see Azañón et al., 2016;Quadt et al., 2018;Park and Blanke, 2019).
As supported by findings across the lifespan and in clinical populations (e.g., Boccia et al., 2020a,b;Raimo et al., 2021a;Raimo et al., 2021b), mental body representations (BR), considering their functional role, can be classified into action-oriented BR (aBR) and nonaction-oriented BR (NaBR, de Vignemont, 2011;Di Vita et al., 2016;Pitron and de Vignemont, 2017).The aBR, or body schema, is a sensorimotor BR involving movements and actions.The NaBR encloses all the other perceptual, conceptual, or emotional BR not finalized for action (e.g., the body structural representation or visuo-spatial body map).
In addition to these functional BR, wich are mainly centered on the outer body, we can also feel and mentally represent the inner body through interoception.Interoception is the sense of the physiological condition of the inner body (Craig, 2002;Garfinkel and Critchley, 2013) and refers to the sensations and representation of physiological signals, spanning from the heartbeat to the itch and to the distressing sensation of not getting enough air (i.e., "air hunger"; Craig, 2002).At the conscious level, interoception can be operationalized along three main dimensions: (i) interoceptive accuracy (IAcc), namely, the performance on objective behavioral tasks, such as heartbeat detection tasks; (ii) interoceptive sensibility (ISe), the self-evaluated tendency to focus on interoceptive signals, gauged using questionnaires; (iii) interoceptive awareness (IAw), the metacognitive awareness of interoceptive accuracy measured through confidence-accuracy correspondence (for an overview, see Garfinkel et al., 2015).
These inner and outer BR are crucial for building up our sense of the self (Craig, 2010;Medina and Coslett, 2016), but also for interacting with others and with the environment (Goldman and de Vignemont, 2009;Dijkerman and Lenggenhager, 2018;Meltzoff and Marshall, 2020;Raimo et al., 2022).For example, imagine going to a pub with some friends; in such a situation, BR play a crucial role in both physical and social interactions.Indeed, in the task of sitting on a chair around a table, the body schema (aBR) interacts with the motor system for planning, executing, and monitoring the actions required for sitting.This involves guiding precise movements needed to position your legs and align the body with the chair.
The visuospatial body map (NaBR), involving the processing of the size of your body parts, boundaries, and spatial relations, could be crucial for estimating the distance between your body and the chair and understanding if you can fit comfortably in that chair.BR also help to calibrate the physical distance from other people in the pub, maintaining a comfortable and appropriate position during social interactions (for an overview of social interactions, and body/peripersonal space representations, see Fanghella, Era and Candidi, 2021).
Moreover, while conversing with your friends, understanding their emotions and desires often requires inferring their internal states.The mechanisms behind these inferences have long been debated but theories of embodied cognition emphasize the crucial role of BR (Singer and Lamm, 2009;Arslanova et al., 2022).Specifically, the inner BR (i.e., interoception) could serve as a blueprint for understanding the experiences of others (Rizzolatti et al., 2006;Singer and Lamm, 2009).In other words, an individual may refer to one's own internal state (e.g., a racing heart when anxious) for comprehending the emotions and desires of others.For example, observing someone in pain might activate the visceral representations of pain in the observer, and this mirror mechanism could be crucial for the empathic response.
Also, the outer BR, for example, in terms of a visuospatial body map (i.e., a representation that contains information about the body borders, the location of body parts, and distance relations between body parts), could be particularly relevant in differentiating between the self and the other, which, in turn, is considered crucial in social interactions for correctly attributing mental and affective states to their origin (Canino et al., 2022).
For this reason, the mental representation of information from the body has been proposed to play a pivotal role in social cognition (Goldman and de Vignemont, 2009;Dijkerman and Lenggenhager, 2018;Gao et al., 2019;Lindblom, 2020).
Social cognition is "the processing of stimuli relevant to understanding agents and their interactions" (Happé et al., 2017).It is a multidimensional construct that embraces core dimensions such as empathy and theory of mind (ToM; also defined as perspective-taking) (Happé et al., 2017), which may be further distinguished into two main dimensions (i.e., affective and cognitive).
More specifically, empathy has been defined as the ability to perceive, understand, and share the emotional experiences of others through both affective/emotional and cognitive processes (Gallese, 2003;Heydrich et al., 2021;Reniers et al., 2011).Affective empathy involves adopting others' emotional experience (i.e., "I feel what you feel"), while cognitive empathy involves understanding others' feeling (i.e., ''I understand how you feel''; Healey and Grossman, 2018;Shamay-Tsoory et al., 2009).Empathy is unique in that it allows an individual to gain insight into someone else's internal state by forming a representation of that state in the observer (Stietz et al., 2019).
Following de Vignemont and Singer (2006), empathy also involves self-other distinction, meaning the awareness that another person is the source of one's emotions.This feature differentiates empathy from emotional contagion, where this awareness is absent.
ToM refers to "the ability to represent one's own mental states and those of others" (Happé et al., 2017;Frith and Frith, 2005;Premack and Woodruff, 1978).According to some neurocognitive models (Healey and Grossman, 2018;Shamay-Tsoory and Aharon-Peretz, 2007;Shamay-Tsoory et al., 2010), ToM can also be dissociable into affective ToM, defined as the ability to infer the emotions or feelings of another individual; and cognitive ToM, defined as the ability to infer the thoughts, beliefs, or intentions of another individual.
Cognitive ToM has been suggested to be a prerequisite for affective ToM, and, in turn, affective ToM involves the cognitive component of empathy and interacts with the affective one (Shamay-Tsoory et al., 2010).
Traditionally, empathy and ToM have been considered and studied as separate processes (see Kanske, 2018;Stietz et al., 2019).Accordingly, different and specific brain areas and networks have been suggested to support empathy and ToM (e.g., Alcalá-López et al., 2018;Kanske et al., 2015; for an overview, see also Stietz et al., 2019).However, this distinction is much less pronounced in real-life social interactions compared to the clear separation observed in controlled experiments.A new promising area of research is currently suggesting that these seemingly independent neural networks do, in fact, interact (Maliske et al., 2023; see also for an overview Maliske and Kanske, 2022).
In line with the idea that cognition is embodied, and specifically that mental representations in various bodily formats could play a crucial role in social cognition (Goldman and de Vignemont, 2009), some studies have empirically demonstrated an association between inner and outer bodily processing and social cognition skills, including empathy (Fukushima et al., 2011;Grynberg and Pollatos, 2015;Raimo et al., 2023) and ToM (e.g., Shah et al., 2017;Canino et al., 2022).Using an interindividual differences approach in healthy individuals, these studies highlight how BR are crucial for interacting with others.A relationship between social cognition, on the one hand, and inner and outer BR, on the other, is also suggested by the literature on some clinical conditions, such as autism and schizophrenia, where BR and social cognition deficits often co-occur (Bird et al., 2010;Shah et al., 2016;Tordjman et al., 2019).Indeed, Tordjman et al. (2019) have proposed a model suggesting that, in schizophrenia and autism, BR disorders would result in problems of self-other differentiation, leading to impaired social cognition.
Similarly, chronic conditions like lower back pain have been associated with altered interoceptive processing and impaired social cognition (Kuner and Flor, 2016).Persistent pain can disrupt the normal interoceptive signals, leading to difficulties in maintaining an accurate perception of internal states, which in turn affects emotional regulation and social interactions (Craig, 2009).
Despite this recent evidence that supports an association between BR and social cognition, results from primary studies are still conflicting.For example, some studies point out that a simple relationship between the inner bodily processing, in terms of IAcc, and specific components of empathy and ToM cannot be predicted (e.g., Ainley et al., 2015).
These mixed findings from individual studies highlight the need for a comprehensive meta-analysis to synthesize the data systematically.However, so far, only a recent review by Baiano et al. (2021) has qualitatively synthesized the findings on the relationship between social cognition ─ in terms of cognitive, affective and visual perspective-taking (with the latter defined by Proulx et al. (2016) as the capacity to understand another agent's visual and spatial experiences) ─ and inner bodily processing, considering different interoceptive dimensions.In their review, Baiano et al. (2021) qualitatively summarized the results from 12 studies, underlining that affective perspective-taking (the Authors' definition of perspective-taking also included empathy) was associated with IAcc.This association highlights that accurately perceiving the body's internal signals could play a crucial role in understanding another person's emotional states (i.e., empathy) with significant real-life implications.For example, enhancing IAcc through specific training could improve understanding of another person's emotions in clinical conditions (e.g., anorexia nervosa, autism spectrum disorders) that show dysfunctions of these abilities (Baiano et al., 2021).However, the results of the association between cognitive perspective-taking and interoceptive dimensions were less clear.Also, Baiano et al. (2021) emphasized the necessity for further research, specifically focusing on ISe and IAw, as these two interoceptive dimensions have been relatively underexplored compared to IAcc.This is particularly relevant because these dimensions are independent (Garfinkel et al., 2015(Garfinkel et al., , 2016) ) and may, therefore, differently affect social cognition.
Thus, despite some crucial insights, a complete picture of the relationship between BR and social cognition is still missing.Indeed, the study by Baiano et al. (2021) was a systematic review and did not include a quantitative summary of findings from various studies through meta-analyses.Moreover, the review by Baiano et al. (2021) only focused on the interoceptive dimensions and did not take into account functional BR (i.e., aBR and NaBR).
Considering that, to the best of our knowledge, no large-scale individual differences studies or meta-analyses have systematically examined the relationships between different BR, including inner and outer BR, and social cognition, we aimed to fill this gap by conducting a comprehensive meta-analysis.

Aims and study plan
Building on a specific interpretation of the embodiment, that is, the idea that BR could play a pivotal role in cognition, including social cognition (Goldman and de Vignemont, 2009), the primary aim of this meta-analysis is to combine and quantitative synthesize results of previous correlational studies in adult populations to clarify the nature and magnitude of the relationship between BR involving the inner and the outer body (i.e., IAcc, ISe, IAw, aBR, and NaBR) and social cognition.In other words, we are interested in understanding if the interindividual variability in BR (despite the reason behind it, that is, sex, age, pathology, and so on) can be associated with interindividual differences in social cognition skills.Only two core social cognition components will be considered: empathy and ToM.
First, we will investigate whether specific relationships exist between different BR and social cognition without differentiating between empathy and ToM.If we find such relationships, considering that empathy and ToM have traditionally been studied as separate processes, we will unpack them further, attempting to determine whether the link may vary for empathy and ToM.
To improve the statistical power of our meta-analysis, more lenient inclusion criteria will be defined, and correlational studies on individuals with clinical conditions will also be included.For this reason, the moderating effect of age, sex and clinical conditions on any observed relationship will be explored.

Study registration
The present meta-analysis was preregistered electronically on the PROSPERO International Prospective Register of Systematic Reviews (CRD42022332294) and was performed following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Page et al., 2021).

Search strategy and study eligibility criteria
A systematic literature search was performed on June 8, 2022, entering the following search terms: ("interocepti* OR interoception OR "body schema" OR "body representation*" OR "body image") AND ("social cognition" OR "theory of mind" OR "perspective taking" OR empathy) using PsychINFO (PROQUEST), PubMed and Scopus databases.The search was not restricted to a specific time frame.
This search was supplemented by examining the reference lists of original and review articles on the same topic, as well as including studies known to be in the process of publication (i.e., studies from our research team).Subsequently, two reviewers (S.C. and V.T.) independently evaluated all papers according to the inclusion and exclusion criteria summarized below.All aspects of study selection, extraction, and assessment were performed by the two reviewers working independently of each other.Disagreements between reviewers were resolved by discussion and consultation with a third researcher.
A primary study was included in the meta-analysis if it: (1) was published in peer-reviewed journals in English; (2) provided results about the correlations between bodily processing (i.e., interoception, aBR or NaBR) with social cognition domains (i.e., ToM, empathy, and their affective and cognitive components) in adult samples (i.e., ≥18yrs.);(3) participants had a normal or correct to normal vision; (4) reported statistical results (i.e., correlation coefficient or statistical significance).
We excluded studies not reporting data about correlations between bodily processing and social cognition components, studies on populations of no interest (i.e., <18 yrs and animal studies), single-case studies, conference proceedings, letters to the editor, theses, commentaries, and reviews as they generally do not provide original data.To evaluate study quality, we employed an adapted version of the Newcastle-Ottawa Scale (NOS; Stang, 2010), a tool commonly used in meta-analyses and systematic reviews to assess the quality of non-randomized studies.The NOS evaluates studies based on three main categories: selection of study groups, comparability of groups, and ascertainment of either the exposure or outcome of interest.Each category has specific criteria, and studies are awarded stars based on how well they meet them.Scores on the NOS range from 0 to 9, with higher scores indicating better methodological quality.In this meta-analysis, the included primary studies have NOS scores ranging from 5 to 7 (see Table 2).These scores indicate moderate to high quality, suggesting that while some studies may have limitations or biases, overall, they meet sufficient criteria to be considered reliable for inclusion in systematic reviews or meta-analyses.
We grouped the questionnaires/tasks under the abovementioned bodily processing dimensions (i.e., interoceptive accuracy; interoceptive sensibility; interoceptive awareness; aBR; NaBR) and social cognition domains (i.e., empathy, ToM, specific affective and cognitive component of empathy; specific affective and cognitive component of ToM; see Table 1), according with the approach reported in Gao et al. (2019) and Goldman and de Vignemont (2009) or with the indication provided in the primary study.When, in a primary study, the same outcome was evaluated using more than one measure, extracted data were reported from the most relevant measure, which was determined by consensus of the two reviewers.

Data extraction and coding
Data extracted and coded from the final articles included: 1. characteristics of the publication (i.e., authors, publication status, year of publication, journal, and country); 2. characteristics of the sample (i.e., total sample size, age, education, sex coded as the percentage of female, disease type, and severity); 3. tests employed to assess interoception, aBR, NaBR, and social cognition (see Table 2).
As for outcome data, the reviewers extracted correlation coefficients or p values of the relationship between interoceptive dimensions, aBR, or NaBR with overall social cognition, and its specific components.When the correlations of our interest or raw data to compute them were not reported in the selected primary studies, we contacted authors; subsequently, in the absence of a response, we used these studies only for qualitative synthesis.Note: For primary research articles including more than one experiment, capital letters were used to differentiate among them; Exp.= Experiment.

Statistical analyses
We synthesized study data using a meta-analytic method, and the statistical analyses were conducted using the software ProMeta 3.0 (2015).To pursue the main aims of the meta-analysis, we calculated the effect sizes (ES) separately from the data reported in the primary studies using Hedges' g unbiased approach.A negative effect size indicates a negative association between constructs of bodily processing and social cognition skills.The conventions used to interpret Hedges'g are similar to those used for Cohen's d.According to Cohen's criteria (Cohen, 2013), values of 0.20, 0.50, and 0.80 represent small, medium, and large effects, respectively.For each ES, we computed the 95% confidence  interval, variance, standard error, and statistical significance.
We then pooled the ES across studies to obtain an overall ES using the inverse-variance method.The random-effects model was used as a conservative approach to account for different sources of variation among studies and to generalize the meta-analytic finding beyond the studies included here.
To assess heterogeneity among studies, we calculated the Q and I 2 statistics indices (Huedo- Medina et al., 2006).A significant Q value indicates a lack of homogeneity among the study findings.The I 2 statistic estimates the proportion of observed variance that reflects real differences in effect sizes, with values of 25 %, 50 %, and 75% considered low, moderate, and high, respectively (Higgins et al., 2003).
To further explain heterogeneity across study findings, we conducted moderator analyses with three moderators ─ age, sex, and disease type ─ assessed through meta-regressions.These moderators were selected based on their potential to influence study outcomes significantly.
We implemented quality control measures, including sensitivity analysis and assessment for publication bias (Bown and Sutton, 2010).Sensitivity analysis involved systematically removing one study at a time from the dataset to assess their impact on between-study heterogeneity and the overall meta-analytic results, thereby testing the robustness of our findings.
The publication bias analysis aimed at revealing biases arising from the tendency of journals to reject negative results (Thornton and Lee, 2000).We applied two methods to explore the publication bias: the funnel plot and the trim and fill procedure.The first is a scatter plot of the ES estimated from individual studies against a measure of their precision (e.g., their standard errors; Rothstein et al., 2005).To evaluate the funnel plot more reliably, we employed Egger's regression method (Egger et al., 1997), which statistically tests the asymmetry of the funnel plot, with nonsignificant results indicative of the absence of publication bias.Instead, the trim and fill procedure (Duval and Tweedie, 2000) is an iterative nonparametric statistical technique for evaluating the effect of potential data censoring on the result of the meta-analyses.In this method, the absence of publication bias is indicated by zero-trimmed studies or, in the presence of trimmed studies, by a trivial difference between the observed and the estimated ES (Duval, 2005).
Studies causing asymmetry in the funnel plot and identified by the trim and fill algorithm were carefully evaluated and, if necessary, excluded from the meta-analysis.

Literature search
Fig. 1 reports the flow diagram based on PRISMA statement.Our search provided 1951 articles after removing duplicates.From these, we selected 246 full-length original articles based on titles and abstracts.After full-text selection, 18 articles were considered eligible.Additionally, three more articles (Canino et al., 2022;Lee et al., 2021;Raimo et al., 2023) were included from other sources such as by inspecting the reference lists cited in the original and review articles on the same topic and studies that we knew were in the process of being published (i.e., studies from our research team).It is also important to note that some of these articles reported more than one experiment.
Of the 21 identified primary research articles: (i) 16 research articles assessed IAcc and social cognition.
(ii) eight research articles assessed ISe and social cognition.
In particular, six articles assessed the relationship between ISe and empathy.Four articles considered both affective and cognitive components of empathy (Mul et al., 2018;Raimo et al., 2023, experiment A and B;Stoica and Depue, 2020;Yang et al., 2021) while two articles did not differentiated between emphatic components (Plans et al., 2021, experiment A and B;Yang et al., 2022).
Three articles assessed the relationship between ISe and ToM, with all three focusing on ISe and affective ToM (Canino et al., 2022;El Grabli et al., 2021;Mul et al., 2018).Two of these articles also assessed ISe and cognitive ToM (Canino et al., 2022;El Grabli et al., 2021).
(iii) 3 articles assessed functional BR and social cognition.
In particular, two articles assessed the relationship between aBR and empathy, both in affective and cognitive components (Lee et al., 2021;Raimo et al., 2023, experiment A and B).One article assessed the relationship between NaBR and empathy, again both in affective and cognitive components (Raimo et al., 2023, experiment B).
Finally, one article assessed the relationship between BR (both aBR and NaBR) and ToM, considering both affective and cognitive components (Canino et al., 2022).
(iv) No studies investigated the relationship between social cognition and interoceptive awareness.
One out of 21 eligible research articles (i.e., El Grabli et al., 2021) did not report the statistical data of interest and was included only in our qualitative synthesis.Similarly, experiment B of the research article by Plans et al. (2021) was included only in our qualitative synthesis since it did not report the statistical data of interest; also, experiment A of this article reported the quantitative data for the relationship between empathy and ISe, but not for the relationship between empathy and IAcc.An additional primary research article (Canino et al., 2022) was included only in the qualitative synthesis since it exclusively explored the associations between variables for which there were not enough studies to obtain sufficient statistical power.
Thus, our final quantitative analyses included 24 experimental studies published in 19 research articles.

Meta-analytic and systematic review results on the association between bodily processing and social cognition
The characteristics of the included studies are summarized in Table 2. Considering the date of publication, the 21 included studies spanned from 2013 to 2023.The present systematic review and metaanalysis enrolled a total of 3466 people, 3314 of them healthy (mean age = 22.18; SD = 3.99), 88 with Autism Spectrum Disorder (mean age = 33.05;SD = 3.82), 36 with Complex Regional Pain Syndrome (mean age = 42.9;SD = 9.8) and 28 with chronic Low Back Pain (mean age =  48.9;SD = 12.2).

Interoceptive sensibility.
The effect size estimated for 8 studies of the association between ISe and social cognition was medium and significant (ES = 0.55, 95 % CI = 0.20-0.91,p = 0.002).Heterogeneity was high (I 2 = 88.31) and significant for the association, and there was neither publication bias (p = 0.301) nor trimmed studies (see Table 3, Fig. 3).

Nonaction-oriented BR.
Since only two studies (i.e., Canino et al., 2022;Raimo et al., 2023, experiment B) investigated the relationship between NaBR and social cognition, we did not perform a   Both studies find a significant association between social cognition and nonaction-oriented Body Representation tasks, suggesting a potential link between social cognition skills and NaBR.

Meta-regression and moderator analysis
The moderation effects of demographic (i.e., age, gender) and clinical (i.e., disease type) aspects of the sample on outcomes were evaluated by conducting meta-regression.According to Borenstein et al. (2009), we conducted a meta-regression for outcomes with at least 10 samples to 1 covariate.Meta-regression revealed that age, sex, and disease type did not significantly affect the association between bodily processing (i.e., interoceptive dimensions, aBR, and NaBR) and social cognition.

Meta-analytic and systematic review results on the association between bodily processing and social cognition components (empathy and ToM)
In this section, we report the results on the association between bodily processing and social cognition components, considering empathy and ToM separately.The characteristics of the included studies are summarized in Table 2.

Interoceptive accuracy and social cognition components
3.4.1.1.Empathy.The effect size estimated for 10 studies of the association between IAcc and empathy was small and marginally significant (ES = 0.28, 95% CI = − 0.03-0.59,p = 0.079).Heterogeneity was high (I 2 = 74.01)and significant for the association, and there was neither publication bias (p = 0.332) nor trimmed studies (see Table 4, Fig. 5).Moreover, an additional article (i.e., Plans et al., 2021) assessed IAcc and empathy, but the statistical data of our interest were unavailable.For this reason, we only qualitatively summarized its results.Plans et al. (2021) suggested that the perception of one's body state was associated with socio-emotional processes, such as empathy.The article included two experimental studies on two different samples.The first study (experiment A) was performed on a sample of 124 healthy participants (age range: 18-63 yrs) and showed higher empathy scores in participants classified as interoceptive using an IAcc task than in participants classified as non-interoceptive.The second study (experiment B) was performed on a sample of 23 healthy participants (age range: 19-53 yrs).In this case, no differences in the empathy scores were found between participants classified as interoceptive and participants classified as non-interoceptive using an IAcc task.

Theory of mind.
The effect size estimated for 9 studies of the association between IAcc and ToM was not significant (ES = 0.16, 95% CI = − 0.08-0.40,p = 0.195).Heterogeneity was moderate (I 2 = 52.37)and significant for the association, and there was neither publication bias (p = 0.184) nor trimmed studies (see Table 4, Fig. 6).Moreover, an additional article (i.e., El Grabli et al., 2021) assessed IAcc and ToM, but the statistical data of our interest were unavailable.For this reason, we only qualitatively summarized its results.El Grabli et al. ( 2021) investigated whether individual interoceptive abilities (IAcc) influenced the ability to understand other people's affective states in a clinical sample with a diagnosis of low back pain (N = 28; mean age: 48.9 yrs; age range: 27-73 yrs).The authors reported no significant correlations between the two constructs in this clinical sample.association between ISe and empathy was large and statistically significant (ES = 0.85, 95% CI = 0.54-1.15,p < 0.001).Heterogeneity was high (I 2 = 81.57)and there was no publication bias (p = 0.975) but the trim and fill analysis removed one study revealing a difference between overall observed ES (0.85) and overall estimated ES (0.72; see Table 4, Fig. 7).

Theory of mind.
Only three studies investigated the relationship between ISe and ToM (Canino et al., 2022;El Grabli et al., 2021;Mul et al., 2018), but statistical data were not available for one study (El Grabli et al., 2021).Thus, since the number of studies with available statistical data was low (n = 2), we did not perform a meta-analysis and summarize the main findings here.The studies involved healthy individuals (Canino et al., 2022), participants with low back pain (El Grabli et al., 2021), and a mixed sample of healthy individuals and those autism spectrum disorder (Mul et al., 2018).The findings from the studies suggest a potential negative relationship between the subjective experience of internal bodily sensations (ISe) and aspects of ToM, particularly in the context of attributing negative emotions to others.However, the conclusion is limited due to inconsistent results across studies.Canino et al. (2022) found a significant negative association between ISe and ToM.While El Grabli et al. ( 2021) and Mul et al. (2018) also observed a negative association between ToM and ISe in clinical samples, these findings were not statistically significant, suggesting variability in results that could stem from differences in sample characteristics, sample size, and the consideration of emotions..4.3. Functional BR and social cognition components 3.4.3.1. Empathy.The effect size estimated for 3 studies of the association between aBR and empathy was absent (ES = 0.00, 95% CI = − 0.19-0.20,p = 0.961).Heterogeneity was absent, and there was neither publication bias (p = 0.952) nor trimmed studies (see Table 4, Fig. 4).

3
Since only one study (i.e., Raimo et al., 2023, experiment B) investigated the relationship between NaBR and empathy, we did not perform a meta-analysis and summarized the main findings here.Raimo et al. (2023) explored the relationship between overall empathy and NaBR in a sample of 164 healthy adults (age range: 19-59 yrs).A weak association between empathy and NaBR was found.

Theory of mind.
Since only one study (i.e., Canino et al., 2022) investigated the relationship between ToM and aBR, and between ToM and NaBR we did not perform a meta-analysis and summarize here the main findings.Canino et al. (2022) explored these relationships in a sample of 82 healthy adults (age range: 18-59 yrs).Concerning the aBR, the authors reported a significant correlation between ToM abilities and performance in action-oriented body representation.This suggests that individuals with a more proficient understanding of their body schema tend to excel in understanding and interpreting their own mental states and that of others.
Concerning the NaBR, the authors reported a significant correlation between ToM abilities and performance in a nonaction-oriented body representation task, suggesting that NaBR plays a role in predicting the individuals' abilities to attribute mental states to others.

Meta-regression and moderator analysis
The moderation effects of demographic (i.e., age, gender) and clinical (i.e., disease type) aspects of the sample on outcomes were evaluated by conducting meta-regression.According to Borenstein et al. (2009), we conducted a meta-regression for outcomes with at least 10 samples to 1 covariate.Meta-regression revealed that age, sex, and disease type did not significantly affect the association between bodily processing (i.e., interoceptive dimensions, aBR, and NaBR) and empathy/ToM.

Meta-analytic and systematic review results on the relationship between bodily processing and the affective and cognitive components of social cognition
A deeper investigation of the relationship between bodily processing and affective/cognitive empathy and ToM components is provided in Supplementary Material 1.In sum, a large effect size was found for the relationship between ISe and cognitive empathy, and a moderate effect size was found for the relationship between ISe and affective empathy.

Discussion
Starting from the idea that cognition is embodied, and specifically that mental representations of the body could play a pivotal role in social cognition (Goldman and de Vignemont, 2009), the present systematic review and meta-analysis aimed to explore the strength of the association between the inner and outer BR and social cognition.
Concerning body processing, a distinction was made between measures that, at the conscious level, probe different interoceptive dimensions (ISe, IAcc, and IAw), and between the higher-order functional BR that can be oriented to action (aBR) or not (NaBR).Concerning social cognition, a distinction was made between measures probing empathy and ToM in their affective and cognitive components.Overall, current findings suggest that specific aspects of the inner and outer BR (i.e., ISe, IAcc and NaBR) are linked to specific aspects of empathy and ToM, advancing our understanding of the effect of body processing on social cognition.

Interoceptive dimensions and social cognition
Regarding internal body processing, we observed a significant effect size for the association between social cognition on the one hand and IAcc and ISe on the other.In particular, when the two core social cognition dimensions (i.e.empathy ad ToM) were considered separately, a large effect size was found for the relationship between the selfevaluated tendency to focus on interoceptive signals (i.e., ISe) and empathy, both for the affective and cognitive components; and a small effect size was found for the relationship between IAcc and empathy.Moreover, no significant relationship was found between ToM and interoceptive dimensions; and in particular between the affective component of ToM and IAcc, while the paucity of studies assessing cognitive ToM and IAcc, as well as ISe and ToM, did not allow us to perform a quantitative synthesis.The lack of studies assessing IAw and social cognition components made it impossible to investigate the strength of their associations.
Our findings expand on the results of Baiano et al. (2021).Indeed, combining the results from different studies through meta-analyses, we mainly found that the interindividual variability in the subjective predisposition to focus on internal bodily information (ISe) was positively associated with both cognitive and affective empathy.In other words, we found that individuals with a heightened self-reported tendency to focus on interoceptive signals, as measured by self-report questionnaires (i.e., ISe), show a higher responsivity to others in terms of a tendency to share (affective empathy) and understand (cognitive empathy) their feelings.Following the shared representations account of social interaction and the idea that the embodied simulation can underpin social cognition (Gallese, 2003(Gallese, , 2007;;Rizzolatti et al., 2006;Singer and Lamm, 2009), we use our own emotion systems in order to feel what another person is feeling.Accordingly, one's own internal (interoceptive) state could serve as a blueprint for understanding and feeling the affective experiences of others.ISe likely represents a trait-like feature; thus, individuals with higher ISe levels could have a tendency to more easily simulate the feelings of an observed person using their own internal state and affective programs, showing higher levels of empathy.This finding supports the neurobiological hypothesis that there are shared brain networks between the intention to feel and understand empathic situations and ISe (see Stoica and Depue, 2020), acknowledging that interoception is linked not only to self-awareness but also to "other-awareness".
However, empathy also involves self-other distinction, that is, the capacity to distinguish between our own affective state and those related to others (Lamm, Bukowski, and Silani, 2016).Individuals with a better ability to accurately detect internal body sensations (i.e., higher IAcc), as mainly measured with heart-beat detection tasks, may be more able to differentiate between themselves and others correctly (e.g., in the rubber hand paradigm, they are less susceptible to body ownership illusions towards the alien body part; Tsakiris et al., 2011;Tajadura-Jiménez and Tsakiris, 2014), resulting, in turn, in higher levels of empathy.Consistently, although this effect was small, we found that the accuracy with which individuals can detect their internal bodily sensations (IAcc) was positively linked with empathy.Indeed, although a shared representation network is suggested to underpin our ability to represent our own feelings and those of others, there is not a complete overlap that would lead to confusion, and successful social interactions also imply distinguishing the representations of the self from that of others (Decety and Sommerville, 2003).Specifically, the ability to differentiate themselves from others and being aware that the other is the source of our affective state is a key feature of empathy that differentiates it from emotional contagion, where this awareness is absent (de Vignemont and Singer, 2006).Also, the self-other distinction is crucial to prevent our own feelings from biasing how accurately we perceive the emotional experiences of others (Lamm, Bukowski, and Silani, 2016).
On the other hand, the lack of a significant relationship between IAcc and ToM could be due to IAcc underpinning low-level mechanisms of social cognition (like those involved in empathy) but not its more sophisticated aspects, like correctly attributing mental states to others.However, additional future studies are necessary to draw definitive conclusions.Indeed, the association between IAcc and cognitive ToM has been underinvestigated, and, concerning the affective ToM, seven out of nine studies included in our meta-analysis employed the Reading the Mind in the Eyes Test (RMET; Baron-Cohen et al., 2001), which requires inferring other individuals' emotional states from a photograph showing only the eye region.Only two studies employed a different ToM measure (i.e., Terasawa et al., 2014;Shah et al., 2017).Specifically, Terasawa et al. (2014) used a task in which the whole face was shown to participants.Shah et al. (2017), instead, evaluated ToM using a more naturalistic measure, that is, the Movie for the Assessment of Social Cognition (MASC; Dziobek et al., 2006) in which the participant has to infer the emotional state of one character watching a short film about a group of 4 individuals at a social event.Interestingly, this study found a significant positive correlation between the performance at the emotional questions of the MASC (i.e., affective ToM) and IAcc.Considering that Dziobek et al. (2006) reported no associations between the MASC and the RMET, suggesting that these tests may assess different aspects of social cognition, future studies with different ToM measures are necessary.Indeed, although speculative, it is possible that interoceptive processing is less relevant for ToM tasks (e.g., the RMET) that rely heavily on very specific external facial features (i.e., the eye region) for appreciating others' emotional states -and where exteroceptive processing (i.e., visual decoding abilities) and the processing of the outer body are possibly more relevant -while it could play a role for affective ToM tasks that do not rely on visual features (e.g., the Emotion Attribution Task; Blair and Cipolotti, 2000) or that involve more naturalistic scenes where the entire face and body of the people are visible, and the voices are audible (e.g., the MASC).
Also, recent findings from a study by von Mohr et al. ( 2021) could suggest a more complex picture of the relationship between IAcc and affective ToM since the timing along the cardiac cycle could modulate it.Indeed, they found that the tendency to use our emotional state when judging the emotional experience of other individuals (i.e., emotional egocentricity bias) varied as a function of the cardiac cycle phase (systole vs. diastole).Specifically, when the other's emotional state was presented at diastole, participants with higher IAcc showed lower emotional egocentricity bias when compared with participants with lower IAcc.However, the opposite pattern was found at systole, that is, an increased emotional egocentricity bias in participants with higher IAcc.In other words, the interaction between the interindividual variability in IAcc and the cardiac cycle phase (systole vs diastole) could be more relevant in shaping specific aspects of social cognition than the interindividual variability in IAcc alone.
Although there were insufficient primary studies concerning the relationship between ISe and ToM to perform a quantitative synthesis, these studies point to a negative relationship between the ability to infer affective mental states and ISe levels; that is, over-reporting interoceptive sensations negatively affects the affective ToM when considering the attribution of negative emotions (Canino et al., 2022).
Thus, ISe could be associated with low-level aspects of social cognition (like those involved in empathy) but not with its more sophisticated aspects, like correctly attributing mental states to others (cognitive ToM), and a heightened ISe could even be detrimental to recognizing other emotions correctly (affective ToM).Although additional studies are necessary to draw definitive conclusions, individuals with higher ISe levels could have the propensity to over-interpret the implied emotion in a described social context (e.g., in performing the Emotion Attribution Task by Blair and Cipolotti, 2000;see Canino et al., 2022) and in others' facial expressions (e.g., could erroneously interpret a neutral face as a face showing fear).Alternatively, challenging social cognition tasks, such as the one involving ToM components, require a shifting of attention from internally to externally focused processes, and thus an excessive focus on interoceptive stimuli that could characterize individuals with higher ISe may be deleterious to the ability to make correct inferences about others' feelings (i.e., affective ToM).Moreover, emotions with a different valence (i.e., positive vs. negative emotions) could show a different pattern of association with ISe, but future studies should better address this assumption.

Higher-order functional BR and social cognition
Finally, regarding the associations with the higher-order functional BR involving the processing also of the outer body (i.e., aBR and NaBR), the lack of studies in the literature allowed us to perform a meta-analysis of only the association between empathy and aBR, with no significant association.However, it should be underlined that in this review, we did not consider a possible additional component of empathy, that is, actionbased or motor empathy (see Blair, 2005), which a study by Raimo et al. (2023) suggests being specifically associated with aBR (or body schema).
There is limited evidence concerning the relationship between NaBR and empathy, although the processing of visuospatial relations and distances among body parts (i.e., NaBR; see Raimo et al., 2023) seems to be associated with the cognitive component of empathy.Individuals with better NaBR could better represent their boundaries and, thus, be more able (as possibly the individuals with higher IAcc) to differentiate between themselves and others correctly, which is an essential step toward understanding what another agent feels (Lamm, Bukowski, and Silani, 2016).This account is consistent with the evidence that in some clinical conditions, such as schizophrenia and autism, body image (i.e., NaBR) and cognitive empathy deficits coexist to the point that Tordjman et al. ( 2019) have suggested a model in which the altered perception of a coherent and unique own body would result in self-consciousness and self-other differentiation problems, leading to impairment in some social cognition components.
Also, the NaBR, along with the aBR, could play a central role in ToM.Indeed, despite the limited evidence in the literature, the study by Canino et al. (2022) showed a significant correlation between cognitive ToM and higher-order BR, which are mainly based on exteroceptive signals.It is interesting to note that in autotopagnosic patients, the damage at the visuospatial body map (i.e., NaBR) results in difficulties in locating body parts that often involve both the own body and others' bodies (for an overview, see Palermo and Di Vita, 2023), implying that the same mechanism is used for representing the visuo-spatial features of one's own body and others' bodies.Similarly, patients with limb apraxia, a neuropsychological syndrome associated with body schema deficits (i.e., aBR; Schwoebel & Coslett, 2005) and characterized by difficulty in performing gestures, can also show deficits in the recognition of the correct execution of familiar gestures performed by others (Pazzaglia et al., 2008).
Thus, efficient NaBR and aBR could be crucial for navigating within these shared representations and simulating what another agent is doing in a social context in order to correctly attribute thoughts, beliefs, or intentions.

Limitations and future directions
Our study is not exempt from limitations, and thus, these findings should be interpreted with caution.
First, although our review provides valuable insights into the relationship between various BR and social cognition, some of our analyses were based on a small number of studies (e.g., three studies with a total sample size of N = 423).While a meta-analysis with few studies still offers significant advantages over relying on individual studies, the small number of included studies certainly limited its statistical power and the generalisability of the findings.
Additionally, there is wide variability between studies in the measures used to evaluate the same construct.To maintain the robustness and validity of the conclusions of the meta-analysis, the present study considered the most frequently employed tests for assessing each domain (e.g., heart-beat tracking task to investigate IAcc).This choice did not reveal whether performances on other tests assessing the same domain (e.g., the heart-beat discrimination task to investigate IAcc) would have resulted in a different association between the interoceptive domains, aBR, NaBR, and social cognition components.Also, we found high levels of heterogeneity that may be due to the employment of different types of tasks.
Finally, the measures to assess the different interoceptive dimensions used in the studies included in this meta-analysis show several limitations that should be taken into account.Indeed, traditional IAcc measures, such as the heart-beat tracking task, suffer from significant psychometric limitations (see for such an account Brener and Ring, 2016).For example, this task appears to have questionable validity since it could reflect the ability of participants to estimate their heartbeats based on previously acquired beliefs rather than the ability to detect heartbeat sensations experienced during the task (Brener and Ring, 2016;Desmedt et al., 2020).Also, concerning the ISe dimension, recent findings suggest that the different ISe questionnaires do not measure a common construct (Desmedt et al., 2022;Vig et al., 2022).For instance, the Body Perception Questionnaire (Porges, 1993) is more focused on the awareness of neutral and negative bodily while the Multidimensional Assessment of Interoceptive Awareness (Mehling et al., 2012) is more focused on an adaptive relationship with body sensations (Desmedt et al., 2022).
Further research is needed to increase the number of evidence in some outcomes (i.e., the relationship between affective and cognitive ToM and NaBR tasks), and further studies investigating all interoceptive dimensions (including the metacognitive dimension) are necessary.We also call for more research exploring social cognition and the possible different patterns of association with the processing of comfortable, neutral, and uncomfortable interoceptive bodily sensations, as well as with different interoceptive submodalities (e.g., the processing of visceral vs. the processing of somatosensory interoceptive information that recent neuroimaging findings in normal and pathological functioning suggest to be underpinned by different neural substrates; see Boccia et al., 2023).Finally, the possibility of nonlinear relationships (e. g., inverted u-shaped relations) between the inner and outer mental representations of the body and social cognition components should be explored.Indeed, for example, higher IAcc and lower IAcc (that could result, respectively, in better/worse self-other distinction and lower/higher self-other overlap) could be equally detrimental for some social cognition tasks that could require a medium amount of self-other distinction and self-other overlap.

Concluding remarks
In conclusion, taken together, our results suggest that the tendency to focus on one's inner sensations is positively related to a more basic aspect of social cognition, such as empathy.In contrast, a higher level of ISe (i.e., over-reporting interoceptive bodily sensations) could be dysfunctional in performing tasks that involve more advanced social cognition dimensions, such as ascribing the mental states of others, especially the affective ones.The kind of embodied self-other distinction that is modulated by IAcc (Tsakiris et al., 2011;Tajadura-Jiménez and Tsakiris, 2014; see also Ainley et al., 2015) could account for the relation between accurate monitoring of the internal body state and empathy, while it could be less relevant for ToM tasks.However, very few studies have investigated the relationship between IAcc and cognitive ToM, while most studies that have investigated the relationship between IAcc and affective ToM have mainly used tasks probing external facial features (i.e., the eye region).
Higher-order functional BR, mainly concerning the outer body processing (such as NaBR), appear to be specifically associated with cognitive ToM, but this needs to be further investigated by other studies (see Fig. 8).
These findings support multi-componential models of social cognition, suggesting a different pattern of association with different kinds of mental representations of the body.
Considering the relevance of outer and inner body processing for cognitive and mental health (Badoud and Tsakiris, 2017;Khalsa et al., 2018;Tsakiris and Critchley, 2016), our systematic review and meta-analytical work may be a starting point for future research focusing on the possible clinical relevance of the association between bodily processing and social cognition.For example, this evidence can be useful in developing evidence-based training to improve social cognition abilities, especially in those individuals with dysfunctional social and empathic skills.
Notes: n = number of participants for each experiment; n F = number of female participants for each experiment; HC = Healthy Controls; ASD: Autism spectrum disorder; cLBP = chronic Low Back Pain; CRPS = Complex Regional Pain Syndrome; NR = Not Reported; IAcc = Interoceptive Accuracy; ISe= Interoceptive Sensibility; aBR = action-oriented body representation; NaBR = nonaction-oriented body representation; ToM = Theory of Mind.a 40 medical students, 20 doctors; 20 non-medical students; 20 non-doctors participants with educational achievements similar to doctors (i.e., academics in other disciplines).

Fig. 4 .
Fig. 4. Forest plot for the relationship between action-oriented BR and social cognition.Notes: ES = effect size; 95% CI = confidence interval; SE = standard error; Sig.= statistical significance; N = total sample size.All the included studies investigated social cognition only in terms of empathy.

Fig. 8 .
Fig. 8.A model of empathy and ToM and the possible associations with ISe, IAcc, aBR, and NaBR.The figure shows a model of empathy and ToM in which the possible associations with ISe, IAcc, aBR, and NaBR are reported.Following Shamay-Tsoorry et al. (2010) and Healey and Grossman (2018), cognitive ToM could be considered a prerequisite for affective ToM; affective ToM involves cognitive empathy.The solid lines indicate associations suggested by the meta-analysis of primary studies; the dashed lines indicate possible associations requiring further investigation.Notes: + positive association; -negative association; aBR (action-oriented body representation); NaBR (nonaction-oriented body representation); IAcc (interoceptive accuracy); ISe (interoceptive sensibility).

Table 1
Classification of tasks/questionnaires tapping the interoceptive domains, actionand nonaction-oriented mental body representations (BR), and social cognition components.

Table 2
Characteristics of the primary studies investigating the relationship between bodily processing and social cognition and classification of the respective investigated domains.

Table 3
Summary of meta-analytic results of studies investigating the relationship between the interoceptive domains, action-and nonaction-oriented mental body representations (BR), and social cognition skills.= number of experiments; N = total number of participants; LL = Lower Limit; UP = Upper Limit; Q and I 2 = heterogeneity statistics; df = degrees of freedom; IAcc = Interoceptive Accuracy; ISe = Interoceptive Sensibility; aBR = action-oriented BR.Exp.= experiment.Statistically significant effect sizes are reported in bold.

Table 4
Summary of meta-analytic results of studies investigating the relationship between the interoceptive domains, action-and nonaction-oriented mental body representations (BR), and empathy/ToM.
Notes: K = number of experiments; N = total number of participants; LL = Lower Limit; UP = Upper Limit; Q and I 2 = heterogeneity statistics; df = degrees of freedom; IAcc = Interoceptive Accuracy; ISe = Interoceptive Sensibility; aBR = action-oriented BR.Exp.= experiment.Statistically significant effect sizes are reported in bold.meta-analysis and summarized here the main findings.Canino et al. (2022) investigated the association between social cognition and body structural representation in a sample of 82 healthy adults (age range: 18-59 years), while Raimo et al. (2023) investigated this association in a sample of 164 healthy adults (age range: 19-59 yrs).